Functional Characterization in Vitro of All Two-component Signal Transduction Systems from Escherichia coli*

Bacteria possess a signal transduction system, referred to as a two-component system, for adaptation to external stimuli. Each two-component system consists of a sensor protein-histidine kinase (HK) and a response regulator (RR), together forming a signal transduction pathway via histidyl-aspartyl phospho-relay. A total of 30 sensor HKs, including as yet uncharacterized putative HKs (BaeS, BasS, CreC, CusS, HydH, RstB, YedV, and YfhK), and a total of 34 RRs, including putative RRs (BaeR, BasR, CreB, CusR, HydG, RstA, YedW, YfhA, YgeK, and YhjB), have been suggested to exist in Escherichia coli. We have purified the carboxyl-terminal catalytic domain of 27 sensor HKs and the full-length protein of all 34 RRs to apparent homogeneity. Self-phosphorylation in vitro was detected for 25 HKs. The rate of self-phosphorylation differed among HKs, whereas the level of phosphorylation was generally co-related with the phosphorylation rate. However, the phosphorylation level was low for ArcB, HydH, NarQ, and NtrB even though the reaction rate was fast, whereas the level was high for the slow phosphorylation species BasS, CheA, and CreC. By using the phosphorylated HKs, we examined trans-phosphorylation in vitro of RRs for all possible combinations. Trans-phosphorylation of presumed cognate RRs by HKs was detected, for the first time, for eight pairs, BaeS-BaeR, BasS-BasR, CreC-CreB, CusS-CusR, HydH-HydG, RstB-RstA, YedV-YedW, and YfhK-YfhA. All trans-phosphorylation took place within less than 1/2 min, but the stability of phosphorylated RRs differed, indicating the involvement of de-phosphorylation control. In addition to the trans-phosphorylation between the cognate pairs, we detected trans-phosphorylation between about 3% of non-cognate HK-RR pairs, raising the possibility that the cross-talk in signal transduction takes place between two-component systems.

The two-component system (TCS) 1 is the signal transduction pathway widely employed from prokaryotes to eukaryotes.
Typically, TCS is composed of a sensor that monitors an external signal(s) and a response regulator that controls gene expression or other physiological activities such as chemotaxis (1). In bacteria, TCS is the major system of signal transduction but not in Archaea and eukaryotes (2). Most of the sensors of bacterial TCS are membrane-associated histidine kinase (HK). The sensor phosphorylates its own conserved His residue in response to a signal(s) in the environment. The carboxyl-terminal cytoplasmic region of HK, called transmitter domain, consists of an ATP-binding domain and a so-called H box domain that includes the conserved His residue for self-phosphorylation. Subsequently, the His-bound phosphoryl group of HK is transferred onto a specific Asp residue on the cognate response regulator (RR) for activation. The activated RR activates, in most cases, transcription of a set of genes, which respond to the external signal.
On the basis of Escherichia coli genome sequence, a total of 30 HK, each containing the conserved self-phosphorylation domain, and a total of 32 RR, each containing the conserved receiver domain, have been predicted (3). After detailed analysis of the genome sequence, two additional RR candidates, YgeK and YhjB, have been identified, both containing the conserved helix-turn-helix motif of the RR family. 2 Recently, Oshima et al. (4) performed a microarray analysis for a total of 30 TCS mutants of E. coli and speculated that, at least for certain combinations, TCSs functionally interact each other to expand the signal transduction network so as to allow some genes to respond to various signals in the environment (4). To examine the specificity of HK-RR interaction in a more direct way, we have purified as many HK and RR proteins as possible, and we tested the self-phosphorylation of HK and the trans-phosphorylation of RR by phosphorylated HK in all possible combinations.

EXPERIMENTAL PROCEDURES
Plasmids-To construct plasmids for overproduction of the cytoplasmic region of each HK containing the HK catalytic domain and the full-length RR, the corresponding DNA fragments were prepared by PCR using E. coli W3110 genome DNA as template and a set of primer pairs (for sequence see supplemental Table I). After digestion of the PCR-amplified fragments with two kinds of the restriction enzyme, each introducing a single cleavage within one of the primer pairs (for sequence see supplemental Table I), the PCR-amplified fragments were inserted into pET21a(ϩ) vector (Novagen) between the same restriction sites as used for the preparation of insert DNAs. All the plasmids thus constructed (supplemental Table II) were confirmed by DNA sequencing.
Protein Expression-To achieve high level expression of the target proteins, each plasmid was transformed into three different competent E. coli cells, two IPTG-inducible strains, BL21(DE3) and JM109(DE3), and one salt-inducible strain BL21(SI) and selected for one strain showing the maximum induction level among the three test strains. For large scale protein production, each transformant was grown in 200 ml of LB broth and at 0.8 -0.9 of A 600 ; IPTG was added to BL21(DE3) and JM109(DE3) cultures at the final concentration of 1 mM, or NaCl was added to BL21(SI) culture at the final concentration of 0.3 M. After 3 h, cells were harvested by centrifugation, washed with lysis buffer (50 mM Tris-HCl, pH 8.0, 4°C, and 100 mM NaCl), and then stored at Ϫ80°C until use.
Protein Purification-Frozen cells were suspended in 3 ml of lysis buffer containing 100 mM phenylmethylsulfonyl fluoride. After addition of 80 l of lysozyme (10 mg/ml), the cell suspension was stored on ice for 30 min and then lysed by sonication. After centrifugation at 15,000 rpm for 20 min at 4°C, the supernatant was mixed with 2 ml of 50% nickel-nitrilotriacetic acid-agarose suspension (Qiagen) and loaded onto a column. After washing with 10 ml of lysis buffer, the column was washed with 20 ml of lysis buffer containing 0.5% Triton X. The Histagged HK or RR protein was then eluted with 2 ml each of lysis buffer containing 0.1, 0.2, or 0.5 M imidazole. The recovery and purity of HK or RR protein in each eluate were checked by SDS-PAGE. The purified HK or RR protein fractions were pooled and dialyzed against storage buffer (50 mM Tris-HCl, pH 7.6, 4°C, 200 mM KCl, 10 mM MgCl 2 , 0.1 mM EDTA, 1 mM dithiothreitol, and 50% glycerol). The protein concentration was determined by the protein assay kit (Bio-Rad), and the purity was checked by SDS-PAGE.
Self-phosphorylation of HK-The purified HK was diluted to 1 M with kinase buffer (50 mM Tris-HCl, pH 8.0, 37°C, 50 mM KCl, and 10 mM MgCl 2 ), and the phosphorylation reaction was initiated by adding 1 Ci of [␥-32 P]ATP at a final concentration of 2.5 M. The reaction was carried out at 37°C for various times and terminated by adding an equal volume of 2ϫ sample buffer (120 mM Tris-HCl, pH 6.8, 4°C, 20% glycerol, 4% SDS, 10% ␤-mercaptoethanol, and 0.1% bromphenol blue). After SDS-PAGE, the gel was washed with 45% methanol, 10% acetic acid, dried, and exposed onto an image plate. The intensity of each band on the image plate was measured with BAS1000 (Fuji Film Co., Japan).
Trans-phosphorylation of RR by HK-The phosphorylated form of HK (1 M) prepared as above was mixed on ice with a mixture of RR (1 M) and excess cold ATP (0.5 mM) and then incubated at 37°C for various times. The reaction was terminated by adding an equal volume of 2ϫ sample buffer. The samples were analyzed by SDS-PAGE. The gel was washed with 45% methanol, 10% acetate, dried, and exposed onto an image plate. The intensity of each gel band was measured with BAS1000 (Fuji Film Co.).

RESULTS
Purification of TCS Components-For in vitro analysis of HK-RR interactions, we tried to purify a total of 64 E. coli TCS components (30 HKs plus HK candidates and 34 RRs plus RR candidates including the newly identified YgeK and YhjB) in His 6 -tagged forms from overexpressed E. coli. The sensor HK is generally composed of two domains, the amino-terminal membrane-associated domain for monitoring an external signal(s) and the carboxyl-terminal cytoplasmic domain with catalytic function of His phosphorylation. Because the signal molecules for HK activation are not yet identified for some sensors, we decided to purify the cytoplasmic HK domain, which alone retains the activity of autophosphorylation and trans-phosphorylation to the cognate RR in the absence of the effector-binding domain (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). The coding sequences for the carboxyl-terminal domain for all 30 sensor HKs or HK candidates with the His 6 tag sequence at the carboxyl terminus were PCR-amplified using sets of primer pairs (for sequence see supplemental Table  I) and were inserted into pET21a(ϩ) to generate the respective expression plasmids (see supplemental Table II). The expression plasmids for 30 RRs and RR candidates were also constructed in the same procedure by using PCR-amplified coding sequences. Expression plasmids for four RRs, CitB, RssB, NtrC, and CreB, were constructed from the respective green fluorescent protein fusion-type Archive clones (16) after digestion with NotI for removal of the green fluorescent protein portion. In these cases, the His 6 tag sequence is added at the amino terminus of each RR coding sequence.
To achieve the maximum level of protein induction, we first checked the expression level by using three different competent strains, IPTG-inducible BL21(DE3) and JM109(DE3) and saltinducible BL21(SI), and under various induction conditions with respect to the inducer concentration and the induction time. The host strain and the best conditions for maximum expression differed from protein to protein (data not shown). Three HK proteins, RcsC, YpdA (B2380), and QseC, were not induced in all the hosts used and under all the conditions tested. Except for these three HK proteins, 61 other proteins (27 HKs and 34 RRs) were subjected to large scale expression and purification. Most of the proteins were recovered in soluble fractions and purified to apparent homogeneity by a single step of affinity chromatography. Two HK proteins (AtoS and YehU) and one RR protein (FimZ) were recovered in pellet fractions after centrifugation and thus solubilized in lysis buffer containing 7 M urea. For these proteins, urea was removed after protein purification by dialyzing against storage buffer without urea. As a result, a total of 61 proteins, 27 truncated form HKs and 34 full-length RR, were purified, all of which apparently showed a single band on SDS-PAGE (data not shown).
Self-phosphorylation of HKs-A total of 27 purified HK was subjected to self-phosphorylation in the presence of radioactive ATP. As shown in Fig. 1, 25 species of the purified HK showed the self-phosphorylation activity, but a detectable level of phosphorylation was not observed for two HKs, CitA and YojN. The YojN protein is not an orthodox-type sensor because it lacks the conserved catalytic domain of His kinase, whereas the lack of self-phosphorylation activity for CitA is not yet clear. For all 25 catalytically active HKs, the self-phosphorylation took place rapidly, and the maximum level of phosphorylation was observed within 30 min. The phosphorylation level did not increase after prolonged incubation for up to 2 h (data not shown). The level of phosphorylation, however, decreased after addition of unlabeled ATP (see below), indicating that the self-phosphorylation of HK is a reversible reaction. The dissociation of radioactive phosphorus was also observed even in the absence of ATP. These observations suggest that the HKs used in this study were de-phosphorylated during purification and after prolonged storage in the absence of ATP.
The rate of the self-phosphorylation activity, shown as the time (t 50% ) required for half of the maximum level, can be classified into four groups (Fig. 2). The R1 group showing the fastest rate of phosphorylation (t 50% , less than 2 min) includes, in decreasing order, ArcB, BasS, NtrB, HydH, PhoR, BarA, PhoQ, and TorS (Fig. 2). The R2 group showing the phosphorylation rate (t 50% ) of 3-5 min includes, in decreasing order, YfhK, NarQ, DcuS, RstB, YedV, and CusS. The R3 group showing the rate of 5-10 min includes, in decreasing order, KdpD, EvgS, UhpB, BaeS, NarX, CheA, CpxA, EnvZ, and CreC. Two R4 group members, AtoS and YehU, showed the lowest rate (t 50% ) of more than 10 min. Both the rate and the level of phosphorylation were very low for AtoC and YehU, supposedly because the contents of active HK molecules were low for these preparations.
Among the 25 HK species analyzed, a total of 17 members showed good correlation between the rate (R) and the level (L) of self-phosphorylation (shaded area in Fig. 2). The activity difference in this group might reflect a difference in the population of active HK molecules in the purified HK preparations. One of the novel findings in this study was the detection of dephosphorylation activity for HK. Even in the presence of [ 32 P]ATP, for instance, the level of phosphorylated NtrB decreased after saturation (see Fig. 1). Five HK members, ArcB (spot A), NtrB (spot Q), HydH (spot M), NarQ (spot O), and EvgS (spot L), showed high rates of phosphorylation but low saturation levels (R Ͼ L, above the shaded area in Fig. 2). As in the case of NtrB, the HKs of this group may have high activities of dephosphorylation under the reaction conditions employed. On the other hand, three HK members, CheA (spot F), BaeS (spot C), and CreC (spot H) (R Ͻ L, below the shaded area in Fig. 2), showed high levels of self-phosphorylation even though the rates of phosphorylation were low.
Trans-phosphorylation of RRs by HKs-From both the previously characterized data of E. coli TCS systems and the paired location of HK and RR genes on the E. coli genome, a total of 26 TCS pairs including both CheA-CheB and CheA-CheY have been predicted (3). The genes for five HKs (arcB, barA, narQ, rcsC, and yojN) and six RRS (arcA, fimZ, narP, rcsB, rssB, and uvrY) are not linked to the genes for respective RR partners on the genome. Based on genetic and/or biochemical data, these orphan HKs and RRs have been considered to form the following cognate HK-RR pairs: ArcB-ArcA, BarA-UvrY, and NarQ-NarP (6,17,18). The RcsC-YojN and YojN-RcsB systems form a sequential pathway, RcsC-YojN-RcsB, of the His-Asp phospho-relay (19). Up to now, however, no HK partners have been identified for two RRs, FimZ and RssB.
By using a total of 25 functional HKs with self-phosphorylation activity (see Figs. 1 and 2), we performed the trans-phosphorylation assay for all possible combinations. A fixed amount of each HK was first incubated with [␥-32 P]ATP for self-phosphorylation until saturation (see Fig. 1), and then mixed with an equal molar amount of the respective cognate RR and an excess amount of unlabeled ATP (100-fold molar excess over the radiolabeled ATP). First we examined the HK-RR cognate pairs. Among a total of 26 cognate HK-RR pairs, including both CheA-CheB and CheA-CheY pairs, a significant level of transphosphorylation was observed at least for 24 pairs (Fig. 3). EvgS failed to phosphorylate EvgA under the reaction conditions employed, whereas trans-phosphorylation level for AtoS- AtoC was hard to detect because these two components could not be separated onto SDS-PAGE.
The level of RR phosphorylation was estimated by measuring both the increase in RR-associated 32 P radioactivity and the decrease in HK-associated radioactivity. The results are summarized in Fig. 4. Both the rate and the level of RR phosphorylation were different among the 24 HK-RR pairs. The maximum level of phosphorylated RR was detected less than 5 min after the addition of phosphorylated HK. The decrease of HKassociated 32 P radioactivity was fast for 16 HK-RR pairs (group A), including pArcB, pBaeS, pCheA, pCpxA, pCreC, pCusS, pDcuS, pKdpD, pNarX, pNarQ, pNtrB, pTorS, pUhpB, pYehU, and pYfhK (where p represents the phosphorylated form). Concomitantly with the dissociation of 32 P radioactivity from HKs, the respective cognate RRs were phosphorylated. The maximum level of trans-phosphorylation in these cases was observed within Ͻ1 min. The lifetime of RR-bound 32 P appears different among RR species. The RR-bound 32 P was stably retained for the A1 group RRs such as pArcA, pCpxR, pCusR, pKdpE, and pNarL but was rapidly released for the A2 group RRs, including pBaeR, pCheB, pCheY, pCreB, pDcuR, pNarP, pNtrC, pTorR, pUhpA, pYehT, and pYfhA. Because the rate of release of RR-bound 32 P was different between the RR species, we concluded that the dephosphorylation rate is an intrinsic property of RR as in the case of HK. Thus, the sum of HK-and RR-bound 32 P showed a time-dependent decrease in most cases, because the dephosphorylation takes place for both phosphorylated HKs and RRs. The rate of trans-phosphorylation was slow for another set of HKs (group B). The cognate RRs, which were phosphorylated by the group B HKs, remained phosphorylated for long periods. After prolonged incubation, the level of phosphorylated RRs increased up to completion for pBasR, pHydG, pOmpR, pPhoB, and pUvrY (group B1). On the other hand, the trans-phosphorylation stopped in the middle of the reaction for pPhoP, pRstA, and pYedW (group B2), and the phosphorylated form of group B2 HKs apparently stayed unchanged at the time 0 level.
Cross-talks in Trans-phosphorylation between Non-cognate HK-RR Pairs-Phosphorylation in vivo of RR by non-cognate HK was suggested by genetic and microarray analysis of some HK mutants (4,17,20). Here we carried out more systematic and direct search for the cross-talk of trans-phosphorylation of RRs by non-cognate HKs. By using 21 species of the functional HK except for TorS and YehU and 34 species of RR, we examined possible cross-talks in trans-phosphorylation for 692 combinations (21 ϫ 34 Ϫ 22 (cognate pairs)). HKs were first selfphosphorylated by incubation with [␥-32 P]ATP for 30 min and then mixed with equimolar amounts of each non-cognate RR and excess unlabeled ATP. After 30 s of incubation, the mixtures were subjected to SDS-PAGE analysis.
Among a total of 692 non-cognate HK-RR pairs, trans-phosphorylation between non-cognate pairs was identified for a total of 22 combinations, as indicated by the red bars in Fig. 5. Seven species of HK (pBarA, pBaeS, pDcuS, pEnvZ, pRstB, pUhpB, and pYedV) phosphorylated non-cognate RR(s). Among these HKs, pUhpB phosphorylated 9 non-cognate RRs and 1 orphan RR (RssB), pBarA phosphorylated 4 non-cognate RRs, and pBaeS phosphorylated 2 non-cognate RRs, whereas pDcuS, pEnvZ, pRstB, and pYedV phosphorylated each non-cognate RR. The external signals sensed by these HKs must regulate, under certain conditions, the genes, which are under the control of another HK-RR system.
On the other hand, nine species of RR (AtoC, CheY, CusR, HydG, KdpE, NarL, NarP, NtrC, and YfhA) were phosphorylated by non-cognate HKs besides their cognate HKs. Genes under the direct control of these RRs must respond to multiple external signals. CusR was phosphorylated by three non-cognate HKs (pBarA, pUhpB, and pYedV), whereas YfhA was phosphorylated by four non-cognate HK (pBaeS, pEnvZ, pRstB, FIG. 2. Relationship between the rate and the level of self-phosphorylation of HK. Self-phosphorylation of the purified HKs was carried out under the standard reaction conditions as described in Fig. 1. The intensity of HKbound 32 P radioactivity was measuring with BAS1000 (Fuji Film Co., Japan), and the net amount of phosphorylated HKs was calculated by using the calibration curve, which provides a linear relation between the phosphorylated HK molecule and the HK-bound 32 P radioactivity. The rate of the self-phosphorylation is shown as the time (t 50% ) required to give half (50%) of the maximum level of phosphorylation. The uppercase letters in each spot corresponds to that in Fig. 1. HKs were classified into four groups (R1 to R4) based on the rate of self-phosphorylation and also into four groups (L1 to L4) on the level of phosphorylation. and pUhpB). CheY, NarL, NarP, and YgeK were phosphorylated by two non-cognate HKs. AtoC was phosphorylated by a non-cognate pUhpB.
Of the two orphan RRs (FimZ and RssB), of which the pairing HK partners have not been identified, RssB was found to be phosphorylated by three HKs, pArcB, pCheA, and pUhpB. RssB binds to the stationary phase-specific RNA polymerase S (RpoS) and transforms it susceptible to degradation by ClpXP proteases. The results described herein suggest that one of the three HKs is the cognate HK of RssB (phosphorylation by other two HKs must be due to cross-talk in trans-phosphorylation). All these HKs are, however, known to form TCS with the respective cognate RR partner (ArcA for pArcB; CheB and CheY for pCheA; and UhpA for pUhpB). Phosphorylation of FimZ was, however, not detected with use of any HKs used in this study.
Interference of TCS Signal Transduction by Non-cognate RR-For some specific combinations, the presence of non-cog-nate RR enhanced dephosphorylation of HK. The enhancement of HK dephosphorylation by non-cognate RR suggests the interference of one TCS signal transduction by another pathway. The combinations of interference are summarized in Fig. 5  (shown by blue lines). Among the RRs tested, five species of RR (CitB, CpxR, PhoB, RssB, and YhjB) enhanced dephosphorylation of more than two species of HK, and four species (CusR, HydG, PhoP, and YfhA) enhanced dephosphorylation of one non-cognate HK. In addition, YhjB, a NarL family orphan RR, stimulated dephosphorylation of 2 HKs, EnvZ and NtrB, even though HK for YhjB phosphorylation was not identified. It is worthwhile to note that RssB, the stability regulator of S (RpoS), interacts with and induces dephosphorylation of 6 HKs, including pBaeS, pCreC, pDcuS, pHydH, pNarQ, pNtrB, and pRstB.
Dephosphorylation of some HKs was enhanced by multiple species of non-cognate RR. For instance, dephosphorylation of pDcuS was enhanced by four non-cognated RRs (CitB, CpxR, PhoB, and RssB) and dephosphorylation of pEnvZ, pNtrB, and pRstB was stimulated each by three non-cognate RRs (Fig. 5). Enhancement of HK dephosphorylation by non-cognate RRs may indicate the interference of signal transduction between different TCSs, leading to expand the cross-talk within the signal transduction network. DISCUSSION TCS is the signal transduction pathway employed in wide varieties of bacteria. Here we carried out, for the first time, a systematic and comprehensive analysis of the activity and specificity of self-phosphorylation in vitro of HK and transphosphorylation in vitro of RR by phosphorylated HK for all purified TCS components from E. coli. For both HK self-phosphorylation and RR trans-phosphorylation, the rate and the level of phosphorylation were found to be different among the HK and RR components. The difference in kinetic parameters of HK self-phosphorylation and trans-phosphorylation may, at least in part, correlate with the nature of each HK and RR such as the need for quick response to changes in environment and/or the duration for maintenance of the memory.
In most cases, the HK with a high rate of self-phosphorylation showed a high level of phosphorylation (see Fig. 2). The activity difference might be related, to a certain extent, to the level of functional protein molecules in HK preparations. However, the amount of phosphorylated form in each HK preparation used might be low, if any, because the HK dephosphorylation takes place during purification and storage. Some HKs such as ArcB, NtrB, HydH, and NarQ showed high rates of phosphorylation but low levels of phosphorylation, presumably because these HKs have the high rate of dephosphorylation ( Fig. 6) as demonstrated for NtrB (see Fig. 1). In contrast, another group including CheA, BaeS, and CreC showed high levels of self-phosphorylation even though the phosphorylation rate was low (see Fig. 6). The HKs of this group may be able to maintain the phosphorylated state (or the response memory) for long periods. The quick response for the HKs of this group, however, must be archived by another step such as the signal sensing of HK, conformational change of membrane-bound HK, and DNA-binding affinity of RR. The difference in kinetic parameters among the test HKs suggests that the in vitro assay of HK self-phosphorylation reflects, to certain extent, the in vivo situations.
Both the rate and level of trans-phosphorylation differed between HK-RR pairs. Based on the rate of trans-phosphorylation, HK-RR pairs could be classified into two groups (see Fig.  4). Group A pairs showed high rates of trans-phosphorylation, but the rate was lower for group B pairs. As in the case of HK self-phosphorylation, the rate and the extent of trans-phosphorylation, herein observed, may reflect the nature of signal FIG. 4. Time course of trans-phosphorylation of RRs by cognate HKs. Trans-phosphorylation of RRs by phosphorylated cognate HKs was carried out for various times as in Fig. 3. The amounts of phosphorylated HKs and RRs at each time point were measured, and the levels relative to the maximum level for each HK or RR are shown. On the basis of kinetic patterns, the reaction profiles are classified into A (subgroup A1 and A2) and B (subgroup B1 and B2) groups (for details see text). Panels a-w are the same pairs of HK and RR as panels a-x in Fig. 3. transduction of each TCS. The group A TCSs are able to respond quickly to changes in the environment (see Fig. 6). It is noteworthy that the three HKs (ArcB, NarQ, and NtrB) with a high rate but a low level of self-phosphorylation are all included in this group.
The group A pairs can be further classified into two subgroups with respect to the stability of phosphorylated RR. The phosphorylated RR is stable for group A1 pairs, whereas group A2 pairs are unstable and rapidly dephosphorylated. The group A1 TCSs are capable of maintaining the response memory for long periods. On the other hand, quick dephosphorylation (or inactivation) of the group A2 RRs may lead to rapid loss of the memory, thereby returning to the steady-state of gene expression after transient activation or repression upon exposure to external stresses (see Fig. 6). The instability of phosphorylated RRs could be intrinsic properties but not due to contamination of nonspecific phosphatases, because the addition of HK or RR of group A2 pairs do not accelerate dephosphorylation of RRs from different TCSs (data not shown). The accurate measurement of the rate and the saturation level of trans-phosphorylation are therefore more difficult than those for selfphosphorylation, because HK carries the phosphatase activity of its own phosphate moiety (this work) and the cognate RR-associated phosphate (21). For instance, the level of pArcB is significantly higher than the input pArcA, suggesting that loss of radioactive phosphate from pArcA prior to its transfer to ArcB.
Our systematic search for trans-phosphorylation of RRs by non-cognate HKs under the same reaction conditions as used for trans-phosphorylation between the cognate pairs detected the cross-talk in phospho-relay for at least 22 combinations (see Fig. 5). In addition, two of the three combinations, which were newly identified for RssB phosphorylation, might be arisen from the cross-talk. Most interestingly, YgeK, a NarL family RR, is unique, because it lacks the receiver domain, which generally contains the conserved Asp residue that is phosphorylated by HK. Most surprisingly, two HKs, pBarA and pUhpB, phosphorylated YgeK (see Fig. 5), suggesting the presence of novel phospho-relay other than the typical His 3 Asp phospho-FIG. 5. Cross-talk in trans-phosphorylation between non-cognate HK-RR pairs. A total of 27 HKs and HK candidates were expressed and purified, except for 3 species (QseC, RcsC and YpdA, shown in green) that were not expressed under the conditions tested. Except for CitA and YojN (shown in blue), self-phosphorylation was detected for a total of 25 HK species (shown in red). Trans-phosphorylation in vitro was then carried out for a total of 850 combinations (25 HKs ϫ 34 RRs). Between the cognate pairs, trans-phosphorylation was detected for 24 pairs (shown by black line) except for two pairs, AtoS-StoC and EvgS-EvgA (shown by black dotted line). Trans-phosphorylations between non-cognate pairs were detected for a total of 19 combinations (shown by red line). Three HKs (ArcB, CheA, and UhpB) phosphorylated RssB, whereas two HKs (BarA and UhpB) phosphorylated YgeK (shown by red dotted line). If each of these two orphan RRs form the cognate TCS pairs with one of these HKs, the rest of the trans-phosphorylation pairs might be attributable to cross-talks, and the total number of cross-talks increases to 22 combinations. Interference of trans-phosphorylation by non-cognate RR was detected for a total of 23 combinations (shown by blue line). Phosphorylation was not detected or not examined for some RRs (shown by black line).